NMR-Based Metabonomic Investigations into the Metabolic Profile of the Senescence-Accelerated Mouse Ning Jiang,†,# Xianzhong Yan,‡,# Wenxia Zhou,*,† Qi Zhang,‡ Hebing Chen,‡ Yongxiang Zhang,† and Xuemin Zhang‡ Beijing Institute of Pharmacology and Toxicology, Beijing 100850, China, and National Center of Biomedical Analysis, Beijing 100850, China Received November 14, 2007
In this work, metabonomic methods utilizing 1H NMR spectroscopy and multivariate statistical technique have been applied to investigate the metabolic profiles of SAM. The serum metabolome of senescenceprone 8 (SAMP8), a murine model of age-related learning and memory deficits and Alzheimer’s disease (AD), was compared with that of control, senescence-resistant 1 (SAMR1), which shows normal aging process. Serum samples were collected for study from both male and female 12-month-old SAMP8 and age matched SAMR1 (n ) 5). 1H NMR spectra of serum were analyzed by pattern recognition using principal components analysis. The results showed that the serum metabolic patterns of SAMP8 and SAMR1 were significantly different due to strains and genders. Subtle differences in the endogenous metabolite profiles in serum between SAMP8 and SAMR1 were observed. The most important metabolite responsible for the strain separation was lack of inosine, which meant the protective function of anti-inflammation, immunomodulation and neuroprotection might be attenuated in SAMP8. Other differential metabolites observed between strains included decreased glucose, PUFA, choline, phosphocholine, HDL, LDL, D-3-hydoxybutyrate, citrate and pyruvate and increased lactate, SFA, alanine, methionine, glutamine and VLDL in serum of SAMP8 compared with those of SAMR1, suggesting perturbed glucose and lipid metabolisms in SAMP8. Besides the differences observed between the strains, an impact of gender on metabolism was also found. The females exhibited larger metabolic deviations than males and these gender differences in SAMP8 were much larger than in SAMR1. Higher levels of VLDL, lactate and amino acids and lower levels of HDL, LDL and unsaturated lipids were detected in female than in male SAMP8. These facts indicated that the metabolism disequilibrium in female and male SAMP8 was different and this may partly explain that females were more prone to AD than males. The results of this work may provide valuable clues to the understanding of the mechanisms of the senile impairment and the pathological changes of AD, as well as show the potential power of the combination of the NMR technique and the pattern recognition method for the analysis of the biochemical changes of certain pathophysiologic conditions. Keywords: senescence-accelerated mouse • metabonomics • 1H NMR spectroscopy
Introduction The senescence-accelerated mouse (SAM) strain has been established as a murine model of accelerated aging through phenotypic selection from a common genetic pool of AKR/J strain of mice by Takeda et al.1,2 SAM is actually a group of related inbred strains including nine strains of senescenceaccelerated-prone (SAMP) and three strains of senescenceaccelerated-resistant (SAMR) mice. The SAMP strains exhibit several features that make them interesting models of human aging, including age-associated early onset of senile amyloidosis, degenerative arthropathy, cataracts, osteoporosis and * To whom correspondence should be addressed. Telephone: 86-10-66931625. Fax: 86-10-6821-1656. E-mail:
[email protected]. † Beijing Institute of Pharmacology and Toxicology. # These authors contributed equally to this work. ‡ National Center of Biomedical Analysis.
3678 Journal of Proteome Research 2008, 7, 3678–3686 Published on Web 07/26/2008
osteoarthritis, reduced fecundity and early loss of fertility.2–4 Among SAMP strains, SAM-prone 8 (SAMP8) shows a shortened life-span and an earlier manifestation of various signs of senescence, including deterioration in learning and memory abilities and various pathological features of neurodegeneration that are observed early in life.5,6 So SAMP8 is a good model to study brain aging and is used as a mouse model of Alzheimer’s disease (AD). Since SAM/resistant 1 (SAMR1) shows a normal aging process, and those deficits and changes in SAMP8 are not observed in SAMR1, it is usually used as a control.7,8 Therefore, the comparison between SAMP8 and SAMR1 might provide clues to understand the basis of the senile impairment and the pathophysiological mechanisms of AD. Various studies have been conducted to assess the agerelated neurochemical alterations and pathological changes in the SAMP8 strain.9–13 In recent years, high-throughput tech10.1021/pr800439b CCC: $40.75
2008 American Chemical Society
research articles
Metabolic Profile of the Senescence-Accelerated Mouse niques including genomics and proteomics have been employed to study the alteration of genes and proteins in SAMP8 mice and the results provide new insights into the mechanism of the cognitive impairment in SAMP8 mice.14–16 For example, it has been found that the altered gene expression and abnormal proteins in SAMP8 mice brain are functionally categorized into neuroprotection, signal transduction, protein folding/degradation, cytoskeleton and transport, immune response and ROS production.15 All of these processes are reportedly involved in age-related cognitive decline.16 However, the changes in genes and proteins are likely to explain only part of the mechanisms that contribute to the cognitive dysfunction in aged SAMP8 mice. Metabolites are the biological end points and the levels of metabolites represent integrative information on cellular function, and consequently define the phenotype of a cell or an organism in response to genetic and environmental changes.17 AD and mild cognitive impairment, a precursor to AD, may be in part due to metabolic insufficiency.18 Therefore, the metabonomic investigations of SAMP8 may provide valuable clues to elucidate the mechanism of AD. There has been increasing application of metabonomics, the quantitative measurement of the multiparametric metabolic response of living systems to perturbation, in complement to genomics and proteomics in the studies of various organisms. One elective technique for metabonomics is nuclear magnetic resonance (NMR) spectroscopy, which allows us to obtain qualitative and quantitative data for many metabolites of the biological system as a whole, in a nondestructive way, without losing the complexity of the systems.19 NMR spectra of sera are represented as complex matrices with several hundreds of proton signals originating from the various metabolites and reflecting the metabolic status of the organism, which alters in response to stressors in order to maintain a homeostatic balance.20 To reduce the interpretational challenge presented by such large data sets, a strategy of data reduction followed by multivariate analysis techniques, such as principal components analysis (PCA), is typically employed. PCA is an exploratory technique of dimension reduction, with each principal component (PC) being a linear combination of the original variables with appropriate weighting coefficients.21 As such, NMR spectroscopy in combination with PCA is powerful for discriminating between samples obtained from abnormal and control animals and investigating the metabolic state of various tissues and fluids in the area of disease diagnosis and drug toxicology.22–24 Serum contains a large variety of low-molecular-weight compounds, which together undergo a variety of possible molecular interactions including metal complexation, chemical exchange processes, micellar compartmentation, enzymemediated biotransformations, and molecular binding.25 Consequently, the metabolic status of the organism can be reflected in the spectral profile and differences between species are observable in the basal NMR spectrum of serum The major advantages of using 1H NMR spectroscopy to study serum metabolites include minimal sample preparation, rapid data acquisition, and the simultaneous provision of diverse information on a variety of low-molecular-weight metabolites and some important macromolecular species. Besides, the high reproducibility of 1H NMR spectra of serum samples is a useful basis for the clinical diagnosis of metabolic and diseased states. While many metabonomic researches on diseases, such as coronary heart disease,24,26 ovarian cancer,27 and other diseases,28 have been reported, few of them were of aging-related
diseases, and the serum metabolic profiles of SAM have not been systematically elucidated. It is therefore important to investigate the metabolic profiles of SAM, for it adds to our basic understanding on the metabolic status of SAM, bestows clues on the relationship between metabolites and functional decline during aging, and establishes baseline data for future metabonomic experiments on the SAM background. In the present study, we investigated the serum profiles of metabolites from male and female SAM to compare the biochemical differences between strains and genders using 1H NMR spectroscopy and pattern recognition (PR) techniques.
Materials and Methods Animal Handling and Sample Collection. SAMR1 and SAMP8 mice were generously provided by Professor T. Takeda (the University of Kyoto, Japan) and transferred to the experimental animal center in our institute in 1994. Mice were housed in rooms with a 12-h light/12-h dark cycle at 20-22 °C with water and food available ad libitum. In this study, male and female SAMR1 and SAMP8 at 12 months of age were used for experiments, 5 for each group. SAMR1 and SAMP8 mice were sacrificed and blood samples were collected in tubes and allowed to clot for 30 min. Sera were obtained by centrifugation (3000 rpm, 10 min at 4 °C) and stored frozen at -70 °C until NMR analysis. During the process of sample collection, two samples from female SAMR1 mice were contaminated accidentally and therefore discarded. Samples Preparation and 1H NMR Measurement. Serum samples were prepared by mixing 100 µL of serum with 350 µL of deuterated water (D2O), which was used as field frequency lock solvent, and 50 µL of 0.1% TSP (3-trimethylsilyl-2H4propionic acid) in D2O added as chemical shift reference (δ 0.0), to a total volume of 550 µL. Samples were centrifuged at 12 000g for 15 min, and the supernatants were transferred into 5 mm NMR tubes. All NMR experiments were carried out on a Varian INOVA 600 spectrometer, operating at proton frequency of 599.97 MHz at 300 K. The data were acquired with 128 scans and a spectral width of 8000 Hz digitized into 64K data points using standard 1D pulse sequence with water presaturation suppression (RD90°-t1-90°-tm-90°-acquire data), resulting in an acquisition time of 4 s. The relaxation delay was set as 2 s and the mixing time was 0.15 s. Saturation of water signal was applied during the relaxation delay (RD) of 2 s and the mixing time (tm) of 150 ms. Serum samples were also analyzed using the Carr-PurcellMeiboom-Gill (CPMG) spin-echo pulse sequence to emphasize the resonances from low molecular weight metabolites. Spectra were acquired with a spin-echo delay τ of 400 µs and a total spin-spin relaxation delay (2nτ) of 100 ms. Other acquisition parameters were the same as described above. A line-broadening factor of 0.5 Hz was applied to data sets before Fourier transformation. In addition, in order to observe the lipid contents of lipoproteins in sera, diffusion-edited experiments were also carried out using a pulse sequence with bipolar pulse pair-longitudinal eddy current delay (BPP-LED).29 The gradient amplitude was set at 35.0 G cm-1 with a length of 1 ms and a diffusion delay of 100 ms. A line-broadening factor of 1 Hz was applied to BPP-LED data sets before Fourier transformation. Data Processing and Multivariate Analysis. All FID data were Fourier transformed, manually phased, and baseline corrected using VNMR software (Varian, Inc.). Only CPMG and BPP-LED data were used for multivariate analysis. For CPMG Journal of Proteome Research • Vol. 7, No. 9, 2008 3679
research articles
Jiang et al.
Figure 1. Low chemical shift regions of 600 MHz 1H NMR CPMG spectra of serum samples from 12-month-old (A) male, (B) female SAMR1, and (C) male, (D) female SAMP8. (3-HB, D-3-hydroxybutyrate; PCho, phosphocholine.)
spectra, each spectrum over the range of δ 0.4 to δ 4.6 was data-reduced into integrated regions of equal width (0.04 ppm). For the data of BPP-LED experiments, each spectrum over the range of δ 0.4 to δ 5.6 was segmented into regions of 0.04 ppm wide and integrated for each region. The region that contained the resonance from residual water (δ 4.6-5.1) was excluded. The integral values of each spectral region was normalized to constant sum of all integrals in a spectrum in order to reduce any significant concentration differences between samples.30,31 Multivariate analysis was carried out by using SIMCA-P+ 10.04 (Umetrics AB, Umeå, Sweden) software package. Principal component analysis (PCA), the most commonly used method for multivariate analysis, or partial least-squares discriminant analysis (PLS-DA), was used to establish the group separation with respect to strain and gender. All variables were preprocessed with mean-centered and Pareto-scaled prior to multivariable analysis. To optimize the clustering of samples purely based on strain, data were also prefiltered by orthogonal signal correction (OSC)32 to remove the confounding variation related to gender and intersubject differences. For this purpose, two dummy Y vectors were created relating to class 1 (SAMR1) and class 2 (SAMP8), consisting of values of 1 for samples belonging to the class and 0 for other samples. Orthogonal component was removed from original data matrix by OSC. Further PLS models were calculated using these filtered data. The results of PCA and PLS were visualized by scores plot of PCs, which display any patterns or groupings within the data and can also be useful in highlighting outliers. The loadings plot reflects the spectral regions responsible for separation in the scores plot.
Results 1
H NMR Spectroscopy of Serum Samples. 1H NMR spectra of biological fluids provide much biochemical information, which is carried in the overall pattern of the metabolic spectral profile. Although NMR spectra produced from sera consist of hundreds of peaks, many of them have been assigned based 3680
Journal of Proteome Research • Vol. 7, No. 9, 2008
on their chemical shifts and signal multiplicities or with the help of two-dimensional correlation spectra.25 The NMR spectra of sera were characterized by a low variability both in metabolite composition and in level, and the peaks exhibited little chemical shift perturbation. The major compounds detected were glucose, lactate, amino acids, and lipids. Figures 1 and 2 show the low and high chemical shift regions of 600 MHz 1H NMR CPMG spectra of serum samples from 12-month-old SAMP8 and SAMR1, respectively. The CPMG pulse sequence edits out those broad resonances and, hence, reveals the smaller molecules in the sample, such as glucose, alanine, and so forth.33 Most of the metabolites were identified and labeled in the spectra, mainly according to the data from literatures.29–31 Visual inspection of these spectra revealed some differences between SAMP8 and SAMR1. The signal intensities of glucose decreased and that of formate increased in the spectra of SAMP8 as compared with those of SAMR1. The most prominent difference between these spectra is the lack of the signals of inosine in SAMP8. In addition, the lipids content of lipoproteins can be observed in BPP-LED spectra33 as shown in Figure 3. The signal intensity of phosphatidylcholine decreased in SAMP8, and the signal from methyl groups of lipids showed different intensities and line shape for SAMP8 and SAMR1, suggesting different contents of lipoproteins in them. HDL and LDL contents (represented by the right half of the methyl signal) in the serum samples of SAMP8 decreased significantly compared with those in SAMR1. These changes are also gender-dependent, revealed by the lower signal intensities of the above-mentioned components in the spectra from females as compared with those from males of the same strain. These reduction were reflected in the decreased intensities of resonant peaks from the methyl groups (δ 0.85) and methylene groups (δ 1.25) of fatty acids, C18 methyl group (δ 0.68) of cholesterol, -N+(CH3)3 of phosphatidylcholine, as well as those from unsaturated fatty acids (UFA) (δ 2.02, 2.78, 5.32).
Metabolic Profile of the Senescence-Accelerated Mouse
research articles
Figure 2. High chemical shift regions of 600 MHz 1H NMR CPMG spectra of serum samples from 12-month-old (A) male, (B) female SAMR1, and (C) male, (D) female SAMP8.
Figure 3. The 600 MHz 1H NMR diffusion-edited spectra of serum samples from 12-month-old (A) male SAMR1, (B) female SAMR1, and (C) male SAMP8, (D) female SAMP8. (PtdCho, phosphatidylcholine.)
Given the high information content and complexity of biological fluids, pattern recognition methods were needed for the detailed interpretation of the NMR spectroscopy data. PCA is a commonly used projection method for overviewing the clustering and trends within data set. The original data are reduced to a few latent variables or principal components (PC) describing maximum variation within the data. The first PC contains the largest proportion of variance in the data set, with subsequent PCs involving progressively smaller proportion of total variance. Therefore, a plot of the first and second PCs may contain a significant proportion of the information content of the original data set.21 The corresponding loadings plots were
examined for the variation in the original variables, describing the influence from the variables on the clustering of samples. Multivariate Analysis of CPMG Data of Serum Samples. Shown in Figure 4 are the PCA results of CPMG data from serum samples of SAMR1 and SAMP8. Before PCA, the data set was preprocessed with Pareto scaling. The scores plot (Figure 4A) shows that PC2 was dominated by strain variation. PC1 and PC2 accounted for 43.3% and 26.3% of total variance (R2X), respectively, with an accumulated Q2 of 51%. In addition, gender-dependent separation was also observed in PC2 for SAMP8, with female SAMP8 located at the extreme end. A PLSDA model for SAMP8 alone resulted in a clear separation Journal of Proteome Research • Vol. 7, No. 9, 2008 3681
research articles
Jiang et al.
Figure 4. (A) PCA scores plot from 1H NMR CPMG spectra of sera obtained from male (purple 2) and female (red b) SAMR1 and male (blue [) and female (black 9) SAMP8. Two samples from female SAMR1 were excluded due to contamination. The first two PCs explained 69.6% of the total variance. (B) 3D PLS-DA scores plot of male (blue [) and female (black 9) SAMP8. R2X ) 79.1%, R2Y ) 98.5%, Q2 ) 93.2%.
Figure 5. PLS scores plot of PC1 vs PC2 (A) and corresponding loadings plot (B) based on the same data set as in Figure 4, but with OSC prefiltering. The first two PCs explained 62.3% of the total variance. See Figure 4 for the explanations of the symbols.
between male and female (Figure 4B), with accumulated R2X of 79.1%, R2Y of 98.5% and Q2 of 93.2%.
(δ 1.48), methionine (δ 2.16), glutamine (δ 2.47) and VLDL (δ 0.91, δ 1.30) increased, as compared with SAMR1.
To emphasize the strain difference, orthogonal signal correction (OSC) was applied to the data sets to remove the residual effect of gender and a PLS model was calculated. Two orthogonal components were removed from the original variables. The first two PCs explained 62.3% of the total variance of X variables and 98.4% of the total variance for Y variables with a Q2 (cum) of 95.5%. The result scores plot (Figure 5A) showed clear separation between SAMP8 and SAMR1 along PC1. The corresponding loadings plot (Figure 5B) revealed the metabolites which contributed to these differences. Detailed analysis on the loadings plot indicated that in SAMP8 the contents of phosphocholine (δ 3.24), HDL and LDL (δ 0.87, δ 1.26), glucose (δ 3.25-3.94), PUFA (δ 2.69-2.85), citrate (δ 2.55, δ 2.70), 3-hydoxybutyrate (δ 1.22) and pyruvate (δ 2.42) decreased, and the contents of lactate (δ 1.34, δ 4.13), alanine
Multivariate Analysis of Diffusion-Edited Spectra of Serum Samples. Diffusion-edited NMR spectra from serum samples were also exploited using PCA to investigate the variation in lipid contents of lipoproteins in different strains. Only the signals from lipid-related parts of lipoproteins were observed in these diffusion-edited spectra (Figure 3). Scores plot of PC1 versus PC3 shows that SAMP8 can be separated from SAMR1 with contributions from both PC1 and PC3 (Figure 6). Gender-dependent distribution is also obvious.
3682
Journal of Proteome Research • Vol. 7, No. 9, 2008
Similar to the CPMG data, PLS model of LED data with OSC prefiltering to eliminate gender effects (two orthogonal component removed) resulted in complete separation between SAMP8 and SAMR1 along PC1 (Figure 7A). Higher VLDL and lower LDL and HDL contents were observed in SAMP8 than in SAMR1 (Figure 7B). In addition, SAMR1 contained much
research articles
Metabolic Profile of the Senescence-Accelerated Mouse 37
Figure 6. PCA scores plot of PC1 vs PC3 based on the diffusionedited 1H NMR spectra of sera from SAMR1 and SAMP8 mice. PC1 and PC3 explained 60.6% and 9.7% of the total variance, respectively. See Figure 4 for explanations of the symbols.
more unsaturated fatty acids and phosphatidylcholine than SAMP8, whereas the signal intensities of R- (δ 2.25) and β-methylene (δ 1.58) groups are higher in SAMP8 than in SAMR1, which may indicate that SAMP8 mice have higher total lipid contents than SAMR1.
Discussion In the postgenomic era, the collective use of global profiling tools, such as genomics, proteomics and metabonomics, are required to fully understand the impact of genetic modifications and toxicological interventions and exposure to stimuli on the network of transcripts, proteins, and metabolites found within cells, tissues, or organisms. These technologies can afford global insight into active cellular processes, without any loss of intrinsic complexity. In particular, metabonomics represents an emerging holistic approach complementary to genomics and proteomics for studying the complex biological system response to chemical and physical input and also to genetic variations.34 In this study, the aging-related serum metabolic profiles of SAMP8 and SAMR1 were holistically investigated using NMR-based metabonomic techniques combined with multivariate statistical analysis. In the present study, we demonstrated that SAMR1 and SAMP8 have different baseline metabolisms in terms of the 1H NMR observable metabolites in serum. In addition to this finding, gender-related differences in some endogenous metabolites were also revealed for SAMP8. However, the dominant trend was the difference between strains rather than gender, that is, strain exerting a greater effect on the metabolism than gender. Metabolic Differences in Serum for Strain Separation. This work has shown a number of substantial metabolic differences in the serum from SAMR1 and SAMP8. The most important metabolites responsible for the strain separation were inosine, glucose, citrate, creatine, D-3-hydroxybutyrate, phosphocholine, polyunsaturated fatty acids, phosphatidylcholine, lactate, methionine, LDL, HDL and VLDL. The largest difference observed between strains was the lack of inosine in SAMP8. Inosine, a major degradation product of adenosine, has been used to relieve the symptoms of many diseases with potent immunomodulatory, neuroprotective and cardioprotective effects.35–39 Inosine can reduce cytokine pro-
duction in inflammatory disease; thus, it is especially beneficial for experimental inflammatory diseases.35,36 In our study, inosine only appeared in serum of SAMR1, but not in SAMP8. This result implied that the protective function of inosine (including anti-inflammation, immunomodulation and neuroprotection) might be attenuated in SAMP8 due to the lack of inosine. Previous studies reported pro-inflammatory cytokines were significantly elevated in 10-month-old SAMP8,40 as well as in AD patients,41,42 and the increases in the expression of pro-inflammatory cytokines may be involved in the agerelated neural dysfunction and/or learning deficiency in SAMP8.40,43 Combining these information, we speculate that the lack of inosine might be one of the reasons for the increased pro-inflammatory cytokines and the immunodeficiency of SAMP8, thus, leading to the learning and memory deficits in SAMP8. The lack of inosine in SAMP8 may also explain the phenomenon that SAMP8 exhibited marked immunodeficiency as early as 2 months after birth, preceding the manifestation of learning deficits.44 Glucose and lipids were two important kinds of metabolites contributing to the significant metabolic differences between the two strains. The levels of glucose, lactate and fatty acids in serum of SAMP8 were significantly different from those of SAMR1. This finding was in agreement with previous reports on SAM, 11,45 which indicated glucose and lipid metabolisms were abnormal in SAMP8. Glucose is the only known substrate from which ATP can be produced in the central nervous system (CNS) and CNS energy production is based almost exclusively upon the oxidation of glucose. Diminished glucose metabolism has been shown to induce the hyperphosphorylation of tau,46 increase production of amyloid beta peptide,47 and influence abnormal neuronal electrical activity. If the down-regulation of energy metabolism related transcripts results in diminution of energy production, this could be expected to impair downstream ATP-dependent processes such as synaptic function and integrity, and ubiquitin-dependent protein degradation pathways, both of which have been shown to be affected in AD.48–51 Lactate is the end product of compounds involved in glycolysis, which is also a potential source of endogenous molecular toxins. Higher levels of lactate were observed in serum of SAMP8 compared to SAMR1 in our study. Such an effect was also observed in severely cognitive-deficient dogs and in early onset AD patients whose lactate production increased 4-fold above controls.52 The increased serum lactate is a key biomarker related to lipoatrophy and lipodystrophy (body fat disorders) and could be consistent with a shift in energy metabolism toward ketone body formation, as evidenced by the increased level of D-3-hydroxybutyrate in SAMP8 observed in this study. The increased lactate level was speculated to reflect the presence of an impaired cerebral oxidative glucose metabolism responsible, at least in part, for its cognitive impairment. The decreased level of glucose and the increased level of lactate in serum suggest that changes in carbohydrate and energy metabolism in SAMP8 occurred. Although there were no significant age associated increases in serum lactate levels in either SAMP8 or SAMR1 mice, the significant decrease in the expression of lactate dehydrogenase-2 in 12 month old SAMP8 mice was observed, 53 which paralleled the observation of elevated serum lactate in SAMP8. Several clinical studies focusing on the plasma glucose metabolism in Alzheimer’s disease found a disproportionate decrease in glucose metabolism in incipient late-onset AD,54 while one of the indicators in early AD was a decrease in glucose metabolism Journal of Proteome Research • Vol. 7, No. 9, 2008 3683
research articles
Jiang et al.
Figure 7. PLS scores plot of PC1 vs PC2 (A) and corresponding loadings plot (B) based on the same data set as in Figure 6, but with OSC prefiltering. The first two PCs explained 62.3% of the total variance. See Figure 4 for explanations of the symbols.
in the parietotemporal association cortex.55 But their results are inconsistent and the association between decreased glucose metabolism and the decline in brain function is still unclear. The major metabolites contributing to lipid metabolism differences between strains were fatty acids and phospholipids. Fatty acids have been found to play an important role in the activity of the central nervous system, which is important in development of cognitive handicap of SAMP8.56 A proper ratio of saturated fatty acid (SFA) and monounsaturated fatty acid (MUFA) is important for membrane fluidity and the change of ratio is related to nervous system disease.57 Unsaturated fatty acid (UFA) is essential for regulating cellular membrane mechanical characteristics and functions and affects the ion permeability. Among UFA, polyunsaturated fatty acid (PUFA) is important in maintaining and improving brain function of aged animals,58 and high PUFA intake could be protective against dementia and predementia syndromes. This effect could be related to the roles of fatty acids in maintaining the structural integrity of neuronal membranes, determining the fluidity of synaptosomal membranes and thereby regulating neuronal transmission59,60 and to the antinflammatory and vascular protection effects of these fatty acids.58,59 Our results showed that the levels of serum SFA in SAMP8 were increased, along with a significant decrease in the levels of UFA. The results, on one hand, indicated that the ratio of SFA and UFA in SAMP8 was in disequilibrium, consistent with reported results,61 and on the other hand, the significantly lower level of PUFA in SAMP8 might lead to the loss of protective effect against dementia due to this PUFA decrease, thus, leading to the learning and memory deficits in SAMP8. Phospholipid metabolism was altered in SAMP8 with the decreased levels of phosphatidylcholine, choline and phosphocholine. Phosphatidylcholine actively contributes to the formation of the pool of free PUFA as a substrate. The synchronous decrease of PUFA and phosphatidylcholine observed in our study supports a relation between phosphatidylcholine and PUFA. Besides, the decreased level of phosphatidylcholine in SAMP8 might imply that cell membrane was destroyed and the formation of the free polyunsaturated fatty acids was impaired too. This suggests that lipid metabolism of cell membrane and membrane-like structures such as mitochondria and endoplasmic reticulum was impaired in SAMP8. In some cases, apoptosis has been correlated with the decrease in phosphatidylcholine level.62,63 On the basis of the above-mentioned facts and our results, we 3684
Journal of Proteome Research • Vol. 7, No. 9, 2008
speculate that apoptosis might occur in SAMP8. In addition, the levels of choline and phosphocholine were also significantly lower in serum of SAMP8 than SAMR1, which was consistent with the finding in cerebrospinal fluid (CSF) of Alzheimer patients.64 The data above suggested that phospholipids metabolism in SAMP8 declined which might be the reason for decreased acetylcholine leading to learning and memory deterioration. Besides inosine, glucose and lipids, lipoproteins and amino acids also contributed to metabolite differences between strains. The abnormal metabolism of lipoproteins was found with lower level of high-density lipoprotein (HDL) and higher level of low-density lipoprotein (VLDL) in SAMP8 compared with SAMR1. Our results showed similar results as those in aged people and it suggested that there existed abnormality of lipoprotein metabolism probably due to the greater lipid or protein oxidation in SAMP8 than in SAMR1.61 However studies on the association between total, LDL, and HDL cholesterol levels and AD have reported divergent results and the relationship of plasma lipid levels to AD risk is far from clear.65 Elevated levels of amino acids, including alanine, methionine, histidine, and phenylalanine, were observed in serum of SAMP8. Serum free amino acid levels corresponded to the protein turnover, including both protein synthesis and protein degradation. The higher levels of free amino acids in serum of SAMP8 were therefore most likely associated with a higher protein turnover and it may also indicate the depressed level of the serum hormone insulin, since insulin can stimulate the absorption of branched-chain amino acids in muscle and fatty tissue synthesis. Previous studies found that damage to the insulin signal transduction cascade may be an early and dramatic event in both pathological conditions, early onset familial Alzheimer disease and late-onset sporadic Alzheimer disease,66 and AD patients with lower insulin levels have lower cognitive skills than those with normal levels.67 Therefore, we suppose that synchronous decreases of the levels of serum free amino acid and glucose found in our study may indicate the occurrence of disturbance of the insulin signal transduction cascade in SAMP8. Metabolic Differences in Serum for Gender Separation. Age and gender are two important factors relating with AD closely. It has been reported that the incidence of AD depended not only on age but also on gender68 and the prevalence of age-related diseases affecting memory, such as AD, is higher
research articles
Metabolic Profile of the Senescence-Accelerated Mouse 69
in females than in males. Similar results were found in SAM that the inheritance of the memory deficit in SAMP8 was either sex-related or -influenced.70 Sex-related metabolic differences in old rats have been observed, but there has been no such report on SAM. In our study, we found that strain differences dominated serum metabolic profiles of SAM, but gender also had an indisputable impact on metabolism to variable extent, especially for SAMP8 (Figure 4B). Our results showed that the metabolic differences between strains are also gender-dependent, that is, the females exhibited larger metabolic deviations than males, as reveled in Figure 4A and Figure 6. In addition, these gender differences in SAMP8 were much larger than in SAMR1. The metabolites relevant to gender differences were the same as those for strain. Higher levels of VLDL, lactate and amino acids and lower levels of HDL, LDL and unsaturated lipids were detected in female than in male SAMP8. Lactate forms the major raw materials for gluconeogenesis and is produced by active skeletal muscle and erythrocytes. Increased lactate level is related to the reduced use of pyruvate in the citric acid cycle and to an increase in anaerobic glycolysis. The increase in lactate level not only suggested that gluconeogenesis was inhibited and changes in carbohydrate and energy metabolism occurred,71 but also reflected the presence of an impaired cerebral oxidative glucose metabolism responsible, at least in part, for cognitive impairment in SAMP8. Furthermore, the higher level of lactate in female SAM implied that glycolysis was more serious in females than males. Lipid metabolism of SAM was different with gender, which was consistent with findings in humans that there was sex dimorphism in postabsorptive fat metabolism in the elderly human72 and aging affects body composition and fuel metabolism differently in each gender, leading to reduced fat oxidation in men, and to an increased ratio of upper- to lowerbody fat in women.73 HDL has long been known to be antiatherogenic,74 and the atheroprotective effect is attributed to the role played by HDL in the reverse transport of cholesterol and to their antioxidant properties.75 The lower level of HDL in female SAM implied females might have greater possibilities of contracting atherosclerosis than male SAM. Our results and the reports that vascular-related diseases, such as diabetes mellitus, hypertension and atherosclerosis, shown to increase the risk of AD, implied that lower level of HDL might contribute to the occurrence of atherosclerosis, thus, increasing the probability of AD risk in female SAM. This speculation was consistent with previous study showing that female SAMP8 tended to have more severely impaired memory function than do males,76 and the differences in the levels of lipids between genders might be one of the reasons for the more severely impaired memory function in female SAMP8. The results we found above indicated that the metabolism disequilibrium of glucose and lipids in female and male SAMP8 was different and suggested that females were more prone to AD than males, which has been described in previous studies.73,74 These results also suggest that in studying the mechanisms of aging, sex-specificity should therefore be considered in animal models.
Conclusion In this paper, the metabonomic analysis based on 1H NMR and multivariate statistical technique has been performed on SAM to study the differences in serum metabolic profiles of SAM. We have demonstrated the strain-related differences in some endogenous metabolites in sera of SAM, along with slight
gender-related effects. The most important metabolites responsible for strain separation were inosine, glucose, lactate, amino acids, fatty acids and lipoproteins, which also contributed to the gender difference. These differences might play important roles in mediating the age-related differences. The dominant trend associated with these metabolic changes was strain, rather than gender. This indicated that the genetic background had a profound effect on the metabolic profiles of SAM. The differential metabolites between strains and genders of SAM found in this study provide new clues to understand the causes of the senile impairment, the pathophysiological mechanisms of AD and the foundation for accelerated senescence in SAM. Moreover, the results in this study clearly show that the NMR spectroscopy-based metabonomic approach is a powerful technique to investigate the biochemical nature of pathophysiological symptoms of complex conditions uninterruptedly and noninvasively. Thereby, we believe that much more clear understanding of the biochemical differences between the senescence-prone and -resistant strains and the discovery of some biomarkers of aging can be achieved, using systematic approaches such as metabonomics, proteomic and genomic techniques. Abbreviations: AD, Alzheimer’s disease; BPP-LED, Bipolar Pulse Pair-Longitudinal Eddy current Delay; CPMG, CarrPurcell-Meiboom-Gill; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MUFA, monounsaturated fatty acid; NMR, nuclear magnetic resonance; NOESY, 1H-1H nuclear Overhauser effect spectroscopy; OSC, orthogonal signal correction; PCA, principal components analysis; PLS, partial leastsquares; PUFA, polyunsaturated fatty acid; PR, pattern recognition; SAM, senescence accelerated mouse; SFA, saturated fatty acid; UFA, unsaturated fatty acid; VLDL, very low-density lipoprotein.
Acknowledgment. This work was supported by grants from the National Natural Science foundation of China (30200367, 90709012, 30701073), the Chinese National Key Project of Basic Research (2004CB518907) and the National High Technology Research and Development Program of China (863 Program, 2007AA02Z400). Note Added after ASAP Publication. This article was published ASAP on July 26, 2008 with an incorrect ref 26. The correct version with ref 26 cited in text was published August 16, 2008. References (1) Takeda, T.; Hosokawa, M.; Takeshita, S.; Irino, M.; Higuchi, K.; Matsushita, T.; Tomita, Y.; Yasuhira, K.; Hamamoto, H.; Shimizu, K.; Ishii, M.; Yamamuro, T. Mech. Ageing Dev. 1981, 17, 183–194. (2) Takeda, T. Neurobiol. Aging 1999, 20, 105–110. (3) Takeda, T.; Hosokawa, M.; Higuchi, K. Exp. Gerontol. 1997, 32, 105–109. (4) Higuchi, K.; Matsumura, A.; Honma, A.; Toda, K.; Takeshita, S.; Matsushita, M.; Yonezu, T.; Hosokawa, M.; Takeda, T. Mech. Ageing Dev. 1984, 26, 311–326. (5) Miyamoto, M.; Kiyota, Y.; Yamazaki, N.; Nagaoka, A.; Matsuo, T.; Nagawa, Y.; Takeda, T. Physiol. Behav. 1986, 38, 399–406. (6) Nomura, Y.; Okuma, Y. Neurobiol. Aging 1999, 20, 111–115. (7) Nomura, Y.; Yamanaka, Y.; Kitamura, Y.; Arima, T.; Ohnuki, T.; Oomura, Y.; Sasaki, K.; Nagashima, K.; Ihara, Y. Ann. N.Y. Acad. Sci. 1996, 786, 410–418. (8) Nie, W.; Zhang, Y. X. Shengli Kexue Jinzhan 2000, 31, 65–68. (9) Nishikawa, T.; Takahashi, J. A.; Fujibayashi, Y.; Fujisawa, H.; Zhu, B.; Nishimura, Y.; Ohnishi, K.; Higuchi, K.; Hashimoto, N.; Hosokawa, M. Neurosci. Lett. 1998, 254, 69–72. (10) Armbrecht, H. J.; Boltz, M. A.; Kumar, V. B.; Flood, J. F.; Morley, J. E. Brain Res. 1999, 842, 287–293.
Journal of Proteome Research • Vol. 7, No. 9, 2008 3685
research articles (11) Shimano, Y. Hokkaido Igaku Zasshi 1998, 73, 557–569. (12) Flood, J. F.; Farr, S. A.; Uezu, K.; Morley, J. E. Mech. Ageing Dev. 1998, 105, 173–188. (13) Morley, J. E.; Farr, S. A.; Kumar, V. B.; Banks, W. A. Peptides 2002, 23, 589–599. (14) Poon, H. F.; Farr, S. A.; Banks, W. A.; Pierce, W. M.; Klein, J. B.; Morley, J. E.; Butterfield, D. A. Brain Res. Mol. Brain Res. 2005, 138 (1), 8–16. (15) Butterfield, D. A.; Poon, H. F. Exp. Gerontol. 2005, 40, 774–783. (16) Poon, H. F.; Castegna, A.; Farr, S. A.; Thongboonkerd, V.; Lynn, B. C.; Banks, W. A.; Morley, J. E.; Klein, J. B.; Butterfield, D. A. Neuroscience 2004, 126, 915–926. (17) Villas-Boas, S. G.; Mas, S.; Akesson, M.; Smedsgaard, J.; Nielsen, J. Mass. Spectrom. Rev. 2005, 24, 613–646. (18) Heininger, K. Rev. Neurosci. 2000, 11, 213–328. (19) Lindon, J. C.; Holmes, E.; Bollard, M. E.; Stanley, E. G.; Nicholson, J. K. Biomarkers 2004, 9, 1–31. (20) Nicholson, J. K.; Buckingham, M. J.; Sadler, P. J Biochem. J. 1983, 211, 605–615. (21) Nicholson, J. K.; Lindon, J. C.; Holmes, E. Xenobiotica 1999, 29, 1181–1189. (22) Beckwith-Hall, B. M.; Thompson, N. A.; Nicholson, J. K.; Lindon, J. C.; Holmes, E. Analyst 2003, 128, 814–818. (23) Bollard, M. E.; Garrod, S.; Holmes, E.; Lindon, J. C.; Humpfer, E.; Spraul, M.; Nicholson, J. K. Magn. Reson. Med. 2000, 44, 201–207. (24) Brindle, J. T.; Antti, H.; Holmes, E.; Tranter, G.; Nicholson, J. K.; Bethell, H. W.; Clarke, S.; Schofield, P. M.; McKilligin, E.; Mosedale, D. E.; Grainger, D. J. Nat. Med. 2002, 8, 1439–1444. (25) Nicholson, J. K.; Buckingham, M. J.; Sadler, P. J. Biochem. J. 1983, 211, 605–615. (26) Holmes, E.; Loo, R. L.; Stamler, J.; Bictash, M.; Yap, I. K.; Chan, Q.; Ebbels, T.; De Iorio, M.; Brown, I. J.; Veselkov, K. A.; Daviglus, M. L.; Kesteloot, H; Ueshima, H.; Zhao, L.; Nicholson, J. K.; Elliott, P. Nature 2008, 453, 396–400. (27) Odunsi, K.; Wollman, R. M.; Ambrosone, C. B.; Hutson, A.; McCann, S. E.; Tammela, J.; Geisler, J. P.; Miller, G.; Sellers, T.; Cliby, W.; Qian, F.; Keitz, B.; Intengan, M.; Lele, S.; Alderfer, J. L. Int. J. Cancer 2005, 113, 782–788. (28) Shockcor, J. P.; Holmes, E. Curr. Top. Med. Chem. 2002, 2, 35–51. (29) Beckwith-Hall, B. M.; Thompson, N. A.; Nicholson, J. K.; Lindon, J. C.; Holmes, E. Analyst 2003, 128, 814–818. (30) Holmes, E.; Nicholls, A. W.; Lindon, J. C.; Ramos, S.; Spraul, M.; Neidig, P.; Connor, S. C.; Connelly, J.; Damment, S. J. P.; Haselden, J.; Nicholson, J. K. NMR Biomed. 1998, 11, 235–244. (31) Waters, N. J.; Holmes, E.; Williams, A.; Waterfield, C. J.; Farrant, R. D.; Nicholson, J. K. Chem. Res. Toxicol. 2001, 14, 1401–1412. (32) Gavaghan, C. L.; Wilson, I. D.; Nicholson, J. K. FEBS Lett. 2002, 530, 191–196. (33) Deprez, S.; Sweatman, B. C.; Connor, S. C; Haselden, J. N.; Waterfield, C. J. J. Pharm. Biomed. Anal. 2002, 30, 1297–1310. (34) Nicholson, J. K.; Connelly, J.; Lindon, J. C.; Holmes, E. Nat. Rev. Drug Discovery 2002, 1, 153–161. (35) Hasko´, G; Sitkovsky, M. V.; Szabo´, C. Trends Pharmacol. Sci. 2004, 25, 152–157. (36) Liaudet, L.; Mabley, J. G.; Pacher, P.; Vira´g, L.; Soriano, F. G.; Marton, A.; Hasko´, G.; Deitch, E. A.; Szabo´, C. Ann. Surg. 2002, 235, 568–578. (37) Mabley, J. G.; Pacher, P.; Liaudet, L.; Soriano, F. G.; Hasko´, G.; Marton, A.; Szabo, C.; Salzman, A. L Am. J. Physiol. Gastrointest. Liver Physiol. 2003, 284, G138–144. (38) Chen, P.; Goldberg, D. E.; Kolb, B.; Lanser, M; Benowitz, L. I. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 9031–9036. (39) de Rougemont, D.; Brunner, F. P.; Torhorst, J.; Wunderlich, P. F.; Thiel, G. Nephron 1982, 31, 310–320. (40) Tha, K. K.; Okuma, Y.; Miyazaki, H.; Murayama, T.; Uehara, T.; Hatakeyama, R.; Hayashi, Y.; Nomura, Y. Brain Res. 2000, 885, 25– 31. (41) Blum-Degen, D.; Muller, T.; Kuhn, W.; Gerlach, M.; Przuntek, H.; Riederer, P. Neurosci. Lett. 1995, 202, 17–20. (42) Fillit, H.; Ding, W. H.; Buee, L.; Kalman, J.; Altstiel, L.; Lawlor, B.; Wolf-Klein, G. Neurosci. Lett. 1991, 129, 318–320. (43) Andre´, A.; Chanse´aume, E.; Dumusois, C.; Cabaret, S.; Berdeaux, O.; Chardigny, J. M. Brain Res. 2006, 1085, 28–32.
3686
Journal of Proteome Research • Vol. 7, No. 9, 2008
Jiang et al. (44) Abe, T.; Yuasa, M.; Kajiwara, Y.; Hosono, M. Cell Immunol. 1994, 157, 59–69. (45) Kurokawa, T.; Sato, E.; Inoue, A.; Ishibashi, S. Neurosci. Lett. 1996, 214, 45–48. (46) Planel, E.; Miyasaka, T.; Launey, T.; Chui, D. H.; Tanemura, K.; Sato, S.; Murayama, O.; Ishiguro, K.; Tatebayashi, Y.; Takashima, A. J. Neurosci. 2004, 24, 2401–2411. (47) Gabuzda, D.; Busciglio, J.; Chen, L. B.; Matsudaira, P.; Yankner, B. A. J. Biol. Chem. 1994, 269, 13623–13628. (48) DeKosky, S. T.; Scheff, S. W. Ann. Neurol. 1990, 27, 457–464. (49) Masliah, E.; Mallory, M.; Alford, M.; DeTeresa, R.; Hansen, L. A., Jr.; Morris, J. C. Neurology 2001, 56, 127–129. (50) Mori, H.; Kondo, J.; Ihara, Y. Science 1987, 235, 1641–1644. (51) Yao, P. J.; Zhua, M.; Pyuna, E. I.; Brooks, A. I.; Therianos, S.; Meyers, V. E.; Coleman, P. D. Neurobiol. Dis. 2003, 12, 91–109. (52) Hoyer, S.; Oesterreich, K.; Wagner, O. J. Neurol. 1988, 235, 143– 148. (53) Poon, H. F.; Castegna, A.; Farr, S. A.; Thongboonkerd, V.; Lynn, B. C.; Banks, W. A.; Morley, J. E.; Klein, J. B.; Butterfield, D. A. Neuroscience 2004, 126, 915–926. (54) Hoyer, S.; Nitsch, R.; Oesterreich, K. J. Neural Transm.: Parkinson’s Dis. Dementia Sect. 1991, 3, 1–14. (55) Heiss, W. D.; Szelies, B.; Kessler, J.; Herholz, K. Ann. N.Y. Acad. Sci. 1991, 640, 65–71. (56) Garbay, B.; Boiron-Sargueil, F.; Shy, M.; Chbihi, T.; Jiang, H.; Kamholz, J.; Cassagne, C. J. Neurochem. 1998, 71, 1719–1726. (57) Ntambi, J. M. J. Lipid Res. 1999, 40, 1549–1558. (58) Solfrizzi, V.; Panza, F.; Capurso, A. J. Neural Transm. 2003, 110, 95–110. (59) Solfrizzi, V.; D’Introno, A.; Colacicco, A. M.; Capurso, C.; Del Parigi, A.; Capurso, S; Gadaleta, A.; Capurso, A.; Panza, F. Exp. Gerontol. 2005, 40, 257–270. (60) Nonaka, N.; Banks, W. A.; Mizushima, H.; Shioda, S. A.; Morley, J. E. Peptides 2002, 23, 2197–2202. (61) Kim, H. C.; Bing, G.; Jhoo, W. K.; Kim, W. K.; Shin, E. J.; Park, E. S.; Choi, Y. S.; Lee, D. W.; Shin, C. Y.; Ryu, J. R.; Ko, K. H. Behav. Brain Res. 2002, 131, 211–220. (62) Al-Saffar, N. M.; Titley, J. C.; Robertson, D.; Clarke, P. A.; Jackson, L. E.; Leach, M. O.; Ronen, S. M. Br. J. Cancer 2002, 86, 963–970. (63) Williams, S. N.; Anthony, M. L.; Brindle, K. M. Magn. Reson. Med. 1998, 40, 411–420. (64) Walter, A.; Korth, U.; Hilgert, M.; Hartmann, J.; Weichel, O.; Hilgert, M.; Fassbender, K.; Schmitt, A.; Klein, J. Neurobiol. Aging 2004, 25, 1299–1303. (65) Michikawa, M. J. Neurosci. Res. 2003, 72, 141–146. (66) Hoyer, S. Eur. J. Pharmacol. 2004, 490, 115–125. (67) Craft, S.; Newcomer, J.; Kanne, S.; Dagogo-Jack, S.; Cryer, P.; Sheline, Y.; Luby, J.; Dagogo-Jack, A.; Alderson, A. Neurobiol. Aging 1996, 17, 123–130. (68) Ishuninaa, T. A.; Swaab, D. F. Exp. Neurol. 2003, 183, 159–172. (69) Andersen, K.; Launer, L. J.; Dewey, M. E.; Letenneur, L.; Ott, A.; Copeland, J. R. M.; Dartigues, J-F.; Kragh-Sorensen, P.; Baldereschi, M.; Brayne, C.; Lobo, A.; Martinez-Lage, J. M.; Stijnen, T.; Hofman, A. Neurology 1999, 53, 1992–1997. (70) Flood, J. F.; Farr, S. A.; Kaiser, F. E.; Morley, J. E. Pharmacol., Biochem. Behav. 1995, 50, 661–664. (71) Solanky, K. S.; Bailey, N. J.; Beckwith-Hall, B. M.; Davis, A.; Bingham, S.; Holmes, E.; Nicholson, J. K.; Cassidy, A. Anal. Biochem. 2003, 323, 197–204. (72) Toth, M. J.; Gardner, A. W.; Arciero, P. J.; Calles-Escandon, J.; Poehlman, E. T. Clin. Sci. 1998, 95, 59–66. (73) Horber, F. F.; Gruber, B.; Thomi, F.; Jensen, E. X.; Jaeger, P. Nutrition 1997, 13, 524–534. (74) Jaouad, L.; de Guise, C.; Berrougui, H.; Cloutier, M.; Isabelle, M.; Fulop, T.; Payette, H.; Khalil, A. Atherosclerosis 2006, 185, 191– 200. (75) Durrington, P. N.; Mackness, B.; Mackness, M. I. Arterioscler. Thromb. Vasc. Biol. 2001, 21, 473–480. (76) Tomobe, K.; Isobe, M.; Okuma, Y.; Kitamura, K.; Oketani, Y.; Nomura, Y. Physiol. Behav. 2005, 84, 505–510.
PR800439B
